61d2a3c36431b3bd433b5db2dbcbd808

This model is a fine-tuned version of Qwen/Qwen2.5-1.5B on the contemmcm/hate-speech-and-offensive-language dataset. It achieves the following results on the evaluation set:

• Loss: 3.0253
• Data Size: 1.0
• Epoch Runtime: 144.0218
• Accuracy: 0.8985
• F1 Macro: 0.6777

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Data Size	Epoch Runtime	Accuracy	F1 Macro
No log	0	0	17.4580	0	11.3268	0.1772	0.1048
No log	1	619	3.8407	0.0078	11.5216	0.6230	0.4599
No log	2	1238	1.6731	0.0156	15.2535	0.8622	0.6961
0.0766	3	1857	1.6686	0.0312	19.8049	0.8876	0.5918
0.0766	4	2476	1.4498	0.0625	26.4385	0.8880	0.6657
1.3466	5	3095	1.2161	0.125	37.8852	0.8914	0.6877
0.1262	6	3714	1.3520	0.25	49.8825	0.8945	0.6032
1.076	7	4333	1.3886	0.5	79.7159	0.8831	0.7345
0.9826	8.0	4952	1.0885	1.0	145.8599	0.9020	0.6863
0.7483	9.0	5571	1.3462	1.0	145.6063	0.8949	0.7303
0.4475	10.0	6190	2.1493	1.0	144.2926	0.9105	0.7205
0.3986	11.0	6809	2.2997	1.0	142.6862	0.9028	0.7125
0.2456	12.0	7428	3.0253	1.0	144.0218	0.8985	0.6777

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.3.0
• Tokenizers 0.22.1